Test for register canceling an unregister.

Service Worker: Tests for unregister then register

These are tests for the Chromium patch r288994:
https://codereview.chromium.org/413063004

Our current shortcomings are:
- New-scope registration overrides the existing registration. This is an
existing bug: https://code.google.com/p/chromium/issues/detail?id=398355
- register() should wait before resolving if docs are using an unregistered
registration. This is a new spec requirement.
- We don't do "early-rejection" described by
https://github.com/slightlyoff/ServiceWorker/issues/396. Also a new spec
requirement.

BUG=388095
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=180141

[falken, 2014-07-28 12:23:23]

Test for https://codereview.chromium.org/413063004/

https://codereview.chromium.org/416003003/diff/60001/LayoutTests/http/tests/s...
File LayoutTests/http/tests/serviceworker/unregister-controller.html (right):

https://codereview.chromium.org/416003003/diff/60001/LayoutTests/http/tests/s...
LayoutTests/http/tests/serviceworker/unregister-controller.html:121: // FIXME:
Failing expectation.
I think this would fail regardless of unregister. We always populate controller
with a new handle in ServiceWorkerProviderHost::SetControllerVersion. This will
have to somehow lookup an existing handle.

[falken, 2014-07-28 12:24:52]

https://codereview.chromium.org/416003003/diff/60001/LayoutTests/http/tests/s...
File LayoutTests/http/tests/serviceworker/unregister-controller.html (right):

https://codereview.chromium.org/416003003/diff/60001/LayoutTests/http/tests/s...
LayoutTests/http/tests/serviceworker/unregister-controller.html:121: // FIXME:
Failing expectation.
On 2014/07/28 12:23:23, falken wrote:
> I think this would fail regardless of unregister. We always populate
controller
> with a new handle in ServiceWorkerProviderHost::SetControllerVersion. This
will
> have to somehow lookup an existing handle.

Actually, my expectation was not right. The windows are different so the workers
won't have the prototype. So we shouldn't expect these to be equal.

[falken, 2014-07-29 02:37:37]

Fixed and simplified the test.

[nhiroki, 2014-07-29 09:06:21]

lgtm

[michaeln, 2014-07-30 02:49:43]

lgtm, but...

i think you should ask jake/alex to look at these tests too, to understand the
detailed consequences of the algos, some of the these outcomes are kinda counter
intuitive to me?

https://codereview.chromium.org/416003003/diff/100001/LayoutTests/http/tests/...
File LayoutTests/http/tests/serviceworker/unregister-then-register.html (right):

https://codereview.chromium.org/416003003/diff/100001/LayoutTests/http/tests/...
LayoutTests/http/tests/serviceworker/unregister-then-register.html:52: // if
registration fails due to a 404 error when fetching the script.
I agree this is the result of the specified algos, but this behavior is probably
going to surprise some people?

// pre - 'something'  is registered at scope

unreg(scope)
reg(bustedthing, scope)

// post - 'something' is still registered at scope

https://codereview.chromium.org/416003003/diff/100001/LayoutTests/http/tests/...
LayoutTests/http/tests/serviceworker/unregister-then-register.html:95: // if
installation of the new worker fails.
ditto

[falken, 2014-07-30 09:41:49]

On 2014/07/30 02:49:43, michaeln wrote:
> i think you should ask jake/alex to look at these tests too, to understand the
> detailed consequences of the algos, some of the these outcomes are kinda
counter
> intuitive to me?
> 

Asked the spec folks here:
https://github.com/slightlyoff/ServiceWorker/issues/396

[falken, 2014-08-07 12:53:50]

michaeln, nhiroki: PTAL. I'd like to revive this patch to assert what we're
currently doing.

It also looks like we're pretty close to how the spec will end up.

https://codereview.chromium.org/416003003/diff/140001/LayoutTests/http/tests/...
File LayoutTests/http/tests/serviceworker/resources/test-helpers.js (right):

https://codereview.chromium.org/416003003/diff/140001/LayoutTests/http/tests/...
LayoutTests/http/tests/serviceworker/resources/test-helpers.js:28:
assert_unreached(error_prefix + ': ' + reason);
This is from asanka's patch: https://codereview.chromium.org/435453002/

[falken, 2014-08-08 05:52:16]

On 2014/08/07 12:53:50, falken wrote:
> michaeln, nhiroki: PTAL. I'd like to revive this patch to assert what we're
> currently doing.
> 
> It also looks like we're pretty close to how the spec will end up.

Actually, hold off on reviewing. The spec just got updated. I'll revive this
patch with the resurrection patch.

[falken, 2014-08-08 07:44:22]

Revised patch to be more in line with current spec. PTAL

Our current shortcomings are:
- New-scope registration overrides the existing registration. This is an
existing bug: https://code.google.com/p/chromium/issues/detail?id=398355
- register() should wait before resolving if docs are using an unregistered
registration. This is a new spec requirement.
- We don't do "early-rejection" described by
https://github.com/slightlyoff/ServiceWorker/issues/396. Also a new spec
requirement.

[falken, 2014-08-11 11:27:47]

On 2014/08/08 07:44:22, falken wrote:
> Revised patch to be more in line with current spec. PTAL
> 
> Our current shortcomings are:
> - New-scope registration overrides the existing registration. This is an
> existing bug: https://code.google.com/p/chromium/issues/detail?id=398355
> - register() should wait before resolving if docs are using an unregistered
> registration. This is a new spec requirement.
> - We don't do "early-rejection" described by
> https://github.com/slightlyoff/ServiceWorker/issues/396. Also a new spec
> requirement.

ping!

[michaeln, 2014-08-12 03:14:28]

lgtm, wdyt about putting the comment below in the CL description?

> Revised patch to be more in line with current spec.
>
> Our current shortcomings are:
> - New-scope registration overrides the existing registration. This is an
> existing bug: https://code.google.com/p/chromium/issues/detail?id=398355
> - register() should wait before resolving if docs are using an unregistered
> registration. This is a new spec requirement.
> - We don't do "early-rejection" described by
> https://github.com/slightlyoff/ServiceWorker/issues/396. Also a new spec
> requirement.

[nhiroki, 2014-08-12 04:58:13]

LGTM2

[falken, 2014-08-13 04:55:04]

The CQ bit was checked by falken@chromium.org

[commit-bot: I haz the power, 2014-08-13 04:55:34]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/falken@chromium.org/416003003/220001

[falken, 2014-08-13 05:14:07]

The CQ bit was unchecked by falken@chromium.org

[falken, 2014-08-13 05:16:50]

On 2014/08/12 03:14:28, michaeln wrote:
> lgtm, wdyt about putting the comment below in the CL description?
> 
> > Revised patch to be more in line with current spec.
> >
> > Our current shortcomings are:
> > - New-scope registration overrides the existing registration. This is an
> > existing bug: https://code.google.com/p/chromium/issues/detail?id=398355
> > - register() should wait before resolving if docs are using an unregistered
> > registration. This is a new spec requirement.
> > - We don't do "early-rejection" described by
> > https://github.com/slightlyoff/ServiceWorker/issues/396. Also a new spec
> > requirement.

Almost forgot this. That's a great idea, done.

[falken, 2014-08-13 05:17:14]

The CQ bit was checked by falken@chromium.org

[falken, 2014-08-13 05:17:44]

The CQ bit was unchecked by falken@chromium.org

[falken, 2014-08-13 05:17:59]

The CQ bit was checked by falken@chromium.org

[commit-bot: I haz the power, 2014-08-13 05:18:13]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/falken@chromium.org/416003003/220001

[commit-bot: I haz the power, 2014-08-13 05:52:45]

Change committed as 180141